JP4953782B2 - Rounding cancellation device - Google Patents

Description

The present invention, OFDM: relates echo cancellation device for canceling the coupling loop interference by (Orthogonal Frequency Division Multiplexing Orthogonal Frequency Division Multiplexing) scheme.

It is known that the OFDM system is not easily affected by delayed waves and is applied to a digital broadcast network using a single frequency, that is, a broadcast wave relay SFN (Single Frequency Network). However, when the broadcast wave relay SFN is realized, the radio wave emitted from the transmission antenna may wrap around the reception antenna and cause problems such as degradation of transmission path characteristics and oscillation of the amplifier.
In general, the following method is employed as a countermeasure against such wraparound. In the first method, first, the frequency characteristic of the sneak path is estimated from the received OFDM signal. Subsequently, the estimated frequency characteristic data of the sneak path is subjected to IFFT (Inverse Fast Fourier Transform) to convert it into impulse response data on the time axis. Next, the impulse response data is set as a filter coefficient in the transversal filter. A method of creating a duplicate signal of a sneak wave, subtracting the duplicate signal from the received signal, and canceling the sneak current is known. The sneak canceller apparatus in this case is configured as shown in FIG. 10, and cancels the sneak wave f2 (for example, Patent Document 1).

  In the second method, first, a scattered pilot (SP) signal is accumulated for a plurality of symbols. Subsequently, the aliasing component is removed by performing Fourier transform with the minimum number of fast discrete Fourier transform (FFT) points. Then, inverse fast discrete Fourier transform (IFFT) is performed. In this way, the frequency characteristic of the sneak path is estimated over the entire channel band, and the frequency characteristic between SP signals is interpolated (for example, Patent Document 2).

  Conventionally, in such a situation, the third and fourth methods have been adopted as a method of taking a countermeasure against wraparound by following the cancel operation. In this method, the cancel operation follows the temporal variation of the wraparound transmission path. Specifically, in the third method, the current wraparound characteristic is predicted from the amount of change in the past filter count. Thereby, even when there is a temporal variation in the sneak wave, a decrease in the tracking characteristic with respect to the variation is suppressed (for example, Patent Document 3).

  The fourth method adaptively controls the prediction coefficient of the prediction circuit to improve the tracking characteristic (for example, Patent Document 4).

Japanese Patent Application No. 2002-98598 Japanese Patent Application No. 2003-343412 Japanese Patent Application No. 2000-156549 (Japanese Patent Laid-Open No. 2001-237749) Japanese Patent Application No. 2003-314736

However, in the methods described in Patent Documents 3 and 4, it has been desired to simplify the calculation of the filter coefficient and improve the followability to the fluctuation of the transmission path.
In the method described in Patent Document 3, prediction by linear combination is performed based on a difference vector between filter coefficients. Then, both the amplitude d and the phase Δθ in FIG. 11 change according to the predicted filter coefficient. For this reason, as shown in FIG. 11, when the fluctuation of the wraparound transmission line rotates concentrically, the predicted filter coefficient w (n) swells outside the circle, that is, from the actual level. Is also predicted to increase in amplitude. Therefore, it is difficult to obtain sufficient performance when applied to an actual propagation path.

In the method described in Patent Document 4, the coefficient of linear combination (weighting coefficient) is adaptively controlled to increase the prediction accuracy. Specifically, paying attention to the fact that the amplitude and phase of each delay wave change every moment, the linear combination coefficient (weighting coefficient) is adaptively controlled for each filter coefficient. As an algorithm for adaptive control, an algorithm that performs processing with a relatively large amount of calculation such as matrix calculation is used. For this reason, a matrix operation is required for each filter coefficient, and the circuit scale becomes relatively large.
Therefore, it has been desired to improve the followability by simple arithmetic processing.

  According to the present invention, followability can be improved by simple arithmetic processing.

(Embodiment 1)
FIG. 1 is a block diagram showing a configuration example of a wraparound canceller apparatus 10 according to Embodiment 1 of the present invention. Here, a case will be described in which the phase rotation amount of the filter coefficient of the FIR filter is compensated at a predetermined update interval. The predetermined update interval may be a plurality of symbol intervals or one symbol interval of a received signal (for example, SP: Scattered Pilot: scattered pilot signal).
In FIG. 1, a wraparound canceller apparatus (hereinafter referred to as a canceller apparatus) 10 includes a reception antenna (reception system) 101, a reception conversion unit 102, an AGC (Auto Gain Control) 103, an A / D (Analog to Digital). ) A conversion unit 104, a subtraction unit 105, an FIR (Finite Impulse Response) filter 106, and a transmission path estimation unit 107. Further, the canceller apparatus 10 includes a sneak estimation unit 108, a coefficient update unit 109, a delay unit 110, a phase rotation unit 111, a D / A (Digital to Analog) conversion unit 112, a transmission conversion unit 113, and a PA (Power Amplifier). ) Part 114 and transmission antenna 115.

The receiving antenna 101 receives a radio frequency band signal (for example, an OFDM signal). Examples of the received signal include a master station wave (input signal) f1 from a master station (not shown) and a sneak wave f2 between the transmitting and receiving antennas 101 and 115. Note that the sneak wave includes delayed waves such as howling and echo.
Reception conversion section 102 amplifies radio frequency band signals (including base station wave f1 and sneak wave f2) received by reception antenna 101. Then, reception conversion section 102 removes unnecessary band components and down-converts them to an intermediate frequency band. For example, an amplifier is used as the amplification method. As the removal method, for example, a band filter is used.
The AGC 103 controls the A / D conversion unit 104 to a level that minimizes quantization noise and noise caused by exceeding the dynamic range. The A / D conversion unit 104 is an A / D converter, for example.

The subtractor 105 receives the signal digitized by the A / D converter 104 and subtracts the cancel signal from this signal. The cancel signal is a signal for canceling the sneak wave f2 between the transmitting and receiving antennas 101 and 115. Thereby, the input signal of the master station wave f1 is taken out.
Then, the subtraction unit 105 outputs the subtracted input signal to the FIR filter (complex FIR filter) 106, the transmission path estimation unit 107, and the D / A conversion unit 112.

The D / A conversion unit 112 converts the output signal of the subtraction unit 105 into an analog signal and outputs the analog signal to the transmission conversion unit 113.
Transmission conversion section 113 up-converts the output signal of D / A conversion section 112 to a radio frequency band and outputs the result to PA section 114. The PA unit 114 amplifies the upconverted signal and radiates it in the air through the transmission antenna 115. As a result, a sneak wave f2 is generated and wraps around the receiving antenna 101.

  The FIR filter 106 receives the output signal of the subtracting unit 105, that is, the signal of the main local wave f1 whose wraparound has been canceled, generates a duplicate of the sneak signal based on the signal, and outputs it to the subtractor 105. The FIR filter 106 updates a filter coefficient (filter tap coefficient value) described later at every predetermined update interval.

The transmission path estimation unit 107 receives the input signal subtracted by the subtraction unit 105 (mainly the signal of the master station wave f1 from which the sneak wave has been canceled), and estimates the characteristics of the sneaking transmission path based on the input signal. .
As this estimation method, a method is generally used in which the frequency characteristic of the transmission path is estimated by conversion to the frequency domain by FFT (see Patent Document 2). The estimation in the transmission path estimation unit 107 is performed at predetermined update intervals.
Then, the transmission path estimation unit 107 outputs the characteristics of the wraparound transmission path to the wraparound estimation unit 108.

The sneak estimation unit 108 estimates the sneak characteristic by obtaining an error between the frequency characteristic of the sneaking transmission path estimated by the transmission path estimation unit 107 and the ideal frequency characteristic without sneaking. The wraparound characteristic represents the output of the subtracting unit 105, that is, the wraparound residual ε (n) of the signal after cancellation. Note that n indicates the output order of residuals.
The estimation in the wraparound estimation unit 108 is performed at predetermined update intervals.
Then, the sneak estimation unit 108 converts the estimated sneak frequency characteristic into a time domain by IFFT, and outputs it to the coefficient update unit 109.

The coefficient updating unit 109 receives the sneak characteristics estimated by the sneak estimation unit 108 and generates and updates the filter coefficients of the FIR filter 106 based on the sneak characteristics.
Specifically, as illustrated in FIG. 2, the coefficient updating unit 109 includes a coefficient multiplying unit 1091, an adding unit 1092, and a delay unit 1093.
The coefficient multiplier 1091 receives the wraparound characteristic ε (n) at the time of update (n-th) from the wraparound estimator 108 and multiplies ε (n) by the update coefficient b. Then, the data is output to the adder 1092. b is a coefficient that cancels ε (n), and satisfies 0 <b ≦ 1. Although b can be controlled dynamically, a fixed value is generally used.

  The adder 1092 adds the output of the coefficient multiplier 1091, that is, bε (n), and the previous (n−1) th filter coefficient c (n−1) acquired from the delay unit 1093. This added value is generated as the updated n-th filter coefficient c (n). Then, the adding unit 1092 outputs the filter coefficient c (n) to the phase rotating unit 111. In this way, filter coefficients are repeatedly generated and updated as new filter coefficients.

Returning to FIG. 1, the phase rotation unit 111 receives the filter coefficient updated by the coefficient update unit 109 and calculates the phase rotation amount of the filter coefficient.
Then, the phase rotation unit 111 calculates the phase rotation amount between the filter coefficient before update and the filter coefficient after update, and causes the FIR filter 106 to set the filter coefficient based on the phase rotation amount. For example, the phase rotation unit 111 sets the calculated phase rotation amount equal to (or otherwise) the phase rotation amount before and after the update, outputs the filter coefficient, and causes the FIR filter 106 to set it. The equality here is not limited to the completely equivalent value, but includes substantially equal values.
When the phase rotation unit 111 is set equal to the phase rotation amount before and after the update, the phase rotation unit 111 acquires the phase rotation amount before the update from the delay unit 110. In the present embodiment, the amount of phase rotation before update is, for example, the past M.
The delay unit 110 accumulates the filter coefficient that is the output of the phase rotation unit 111. The delay unit 110 is configured by, for example, a RAM (Random Access Memory).

The configuration of the phase rotation unit 111 will be described in detail. As shown in FIG. 3, the phase rotation unit 111 includes a vector signal phase rotation unit 1111, a phase rotation amount calculation unit 1112, and an average amount calculation unit 1113.
The vector signal phase rotation unit 1111 rotates the filter coefficient c (n) acquired from the coefficient update unit 119 by the average value of the phase rotation amount acquired from the average amount calculation unit 1113. As a result, the filter coefficient c (n) rotates by a predetermined amount (an average value of the phase rotation amount).
Vector signal phase rotation section 1111 outputs the filter coefficient to FIR filter 106. As a result, the FIR filter 106 inputs a filter coefficient obtained by averaging the phase rotation amounts for the past number of updates, and generates a copy of the wraparound signal.

The phase rotation amount calculation unit 1112 acquires the filter coefficients before and after the update acquired from the coefficient update unit 119 (for example, the past M times stored in a storage unit (not shown)), and calculates the total phase difference of these filter coefficients.
The average amount calculation unit 1113 calculates an average value of all the phase differences calculated by the phase rotation amount calculation unit 1112 (for example, the past M times stored in a storage unit (not shown)). An example of this calculation formula is expressed in Equation 1. Thereby, it is possible to calculate in consideration of increase / decrease of the phase difference from the past.

  It is assumed that the phase rotation unit 111A in FIG. 4 sets the phase rotation amount instead of the phase rotation unit 111 in FIG. The phase rotation unit 111A obtains a phase difference from the filter coefficient at the time of update and the filter coefficient at the previous time, and rotates the filter coefficient at the time of update so as to compensate for the phase difference.

  However, when the phase rotation amount Δθ is set in this way, there is a possibility that a vibration phenomenon occurs. This is because of the following reason. That is, the vector signal phase rotation unit 1111 obtains the phase difference Δθ (n) from the nth filter coefficient at the time of update and the n−1th filter coefficient. Then, in order to compensate for Δθ (n), the phase rotation amount calculation unit 1112 calculates the phase rotation amount by rotating the nth filter coefficient by Δθ (n), and sets the filter coefficient. A formula for calculating the phase rotation amount Δθ at this time is expressed by Formula 2.

Then, at the next (n + 1) th time, the transmission path estimation unit 107 assumes that there is no fluctuation in the transmission path and estimates the characteristics of the transmission path. As a result, the coefficient updating unit 109 outputs the same n + 1-th filter coefficient as the n-th filter coefficient to the phase rotation unit 111A. For this reason, in the phase rotation unit 111A, the filter coefficient is set with the phase difference between the filter coefficients before and after the update (n + 1 and n) as zero. Then, the characteristics of the next (n + 2) -th transmission path are also estimated as if the transmission path fluctuated again, and operate so as to compensate for the phase difference Δθ (n + 1). That is, the phase compensation value vibrates in the form of Δθ (n−1), 0, Δ (n + 1), 0, Δ (n + 3), 0,. Therefore, the reception (relay) quality may not be improved. It is assumed that Δθ (n−1) ≈Δθ (n + 1) ≈Δ (n + 3) is the transmission path variation condition.
Thus, the method of averaging the phase rotation amount in the phase rotation unit 111 in FIG. 3 can also reduce the vibration of the phase compensation value, unlike the case of setting the phase rotation amount in the phase rotation unit 111A in FIG. .

[Other configuration examples of phase rotation unit]
Next, another configuration example of the phase rotation unit will be described with reference to FIG.
The phase rotation unit 111B in FIG. 5 includes the coefficient rotation unit 1114, the addition unit 1115, and the delay unit 1116 in addition to the phase rotation unit 111 in FIG. With this configuration, the phase rotation unit 111B accumulates the phase rotation error and calculates the filter coefficient. Thereby, it becomes possible to make the followability to the fluctuation | variation of a transmission path more highly accurate.
The coefficient multiplication unit 1114 multiplies the phase rotation amount Δθ (n), which is the output of the phase rotation amount calculation unit 1112, by a coefficient α. α satisfies 0 <α <1. In particular, when it is assumed that the transmission path is capable of canceling the wraparound, α preferably satisfies 0 <α ≦ 0.5.

  The adder 1115 outputs αΔθ (n) that is the output of the coefficient multiplier 1114 and the previous accumulated value {αΔθ (n−1) + αΔθ (n−2) +. Addition and output to the vector signal phase rotation unit 1111. Then, the vector signal phase rotation unit 1111 converts the filter coefficient c (n) acquired from the coefficient update unit 119 into the phase rotation amount {αΔθ (n−1) + αΔθ (n) + αΔθ (n−) that is the output of the addition unit 1115. 2) Rotate +...} And output to the FIR filter 106.

  Thus, the FIR filter 106 inputs the filter coefficient c (n) having the phase rotation amount Δθ = {αΔθ (n−1) + αΔθ (n) + αΔθ (n−2) +. Will be generated. Therefore, the error can be smoothed, and the followability to the transmission path is improved.

[Combination example of phase rotation unit and coefficient update unit]
Here, the phase rotation unit 111B of FIG. 5 may be combined with the coefficient update unit 109A shown in FIG. By this combination, the affinity is improved in terms of followability with respect to fluctuations in the transmission path, which is preferable.
The coefficient update unit 109A in FIG. 6 is configured by removing the delay unit 1093 of the coefficient update unit 109 shown in FIG. That is, the coefficient update unit 109A includes a coefficient multiplication unit 1091 and an addition unit 1092. Then, the adder 1092 adds the output of the coefficient multiplier 1091 and the previous filter coefficient obtained from the delay unit 110. As a result, the output value of the adder 1092 in FIG. 6 is the same as the output value of the adder 1092 in FIG. The configuration of the sneak device including the other phase rotation unit 111B is the same as in the case of FIGS. With this configuration, the delay unit 1093 in FIG. 2 is not necessary, and the circuit scale can be reduced.

  Note that the delay unit 110 is required to correspond to each tap of the FIR filter 106. In order to satisfy the performance required as a wraparound canceller, the number of taps corresponding to the guard interval length is required. For example, in the case of digital terrestrial broadcasting mode 3 and the guard interval length is 1/8 of the effective symbol, about 1024 taps are required. For this reason, if it comprises RAM etc., the circuit scale of the delay part 110 can be made smaller.

As described above, according to the present embodiment, the sneak signal for canceling the sneak current between the transmitting and receiving antennas 101 and 115 is subtracted from the signal from the reception antenna 101 (master station wave f1 and sneak wave f2). Subsequently, based on the input signal after subtraction, a duplicate of the sneak signal according to the filter coefficient of the FIR filter 106 is generated. Next, the characteristic of the sneaking transmission path is estimated based on the input signal after subtraction, and the sneaking characteristic is estimated based on the characteristic of the sneaking transmission path. Further, the filter coefficient is generated and updated based on the wraparound characteristic, the phase rotation amount of the filter coefficient is calculated and made equal to the phase rotation amount before the update, and the filter coefficient is set in the FIR filter 106.
For this reason, the FIR filter 106 is set with a filter coefficient with an equal amount of phase rotation before and after the update, and the followability to the fluctuation of the transmission path is improved. Furthermore, since the amplitude of the filter coefficient is always kept constant, the arithmetic processing is simplified. Therefore, the hardware (circuit) scale of the canceller device can be reduced.

(Embodiment 2)
In the second embodiment, when the filter coefficient update interval is a plurality of symbol intervals (for example, four symbol intervals), the phase rotation amount of the filter coefficient is compensated for each symbol interval (one symbol interval).
First, the operation timing for compensating the phase rotation amount of the filter coefficient in the conventional example will be described (see Patent Document 3 and the first embodiment).
FIG. 7 is a diagram showing the operation timing for compensating the phase rotation amount of the filter coefficient in the conventional example.
The four patterns A to D in FIG. 7 show the SP signal insertion positions. That is, four patterns A to D are repeated every four symbols. A series of processes from the acquisition of the SP signal (“take” in FIG. 7) to the calculation of the phase rotation amount (“calculation” in FIG. 7) until the filter coefficient is updated (“update” in FIG. 7). Is performed every four symbols.
Specifically, the SP signal is captured at the timing when the GI (guide interval) between pattern A and pattern B ends (“capture” in FIG. 7). Then, the phase rotation amount is calculated during the three symbol periods (patterns B to D) ("calculation" in FIG. 7), and as a result, the filter coefficient is updated at the timing when pattern D ends (FIG. 7). Medium "update").

FIG. 8 is a diagram illustrating operation timing for compensating for the phase rotation amount of the filter coefficient in the second embodiment.
According to FIG. 8, the phase rotation amount of the filter coefficient is compensated for each symbol section (one symbol section) even during the update interval of four symbol sections, unlike the case of FIG. In this case, the magnitude of the phase rotation compensation value is a multiple of a value obtained by dividing the phase rotation amount obtained between update intervals by (number of update interval symbols).
For example, when the phase rotation amount of pattern A obtained at the n-th update timing is Δθ (n), the filter coefficient of the next symbol of pattern B is rotated by {Δθ (n)} / 4. Then, the filter coefficient of the symbol of the next patternC is rotated {2Δθ (n)} / 4. Further, the filter coefficient of the symbol of the next patternD is rotated by {3Δθ (n)} / 4. In this way, the phase rotation amount of the filter coefficient is compensated for each symbol section (one symbol section).

Next, a configuration example of the phase rotation unit 111D that realizes such phase rotation compensation will be described.
FIG. 9 is a block diagram illustrating a configuration example of the phase rotation unit 111D of the second embodiment.
The phase rotation unit 111C in FIG. 9 is configured to further include a division unit 1117 and a multiplication unit 1118 in addition to the phase rotation unit 111B in FIG. The configuration of the canceller apparatus including the other phase rotation units is the same as in the case of FIGS.
The division unit 1117 calculates a phase rotation amount per symbol from the phase rotation amount (for the update interval) that is the output of the addition unit 1115. This calculation formula is the amount of phase rotation Δθ / M. M represents the number of symbols in the update interval.

Multiplier 1118 multiplies the phase rotation amount output from division unit 1117 by the phase rotation amount coefficient, and outputs the result to vector signal phase rotation unit 1111. As a result, the vector phase rotation unit 1111 rotates the phase rotation amount that is the output of the multiplication unit 1118.
At this time, in the present embodiment, since the update interval is a 4-symbol section (M = 4), the phase rotation amount coefficient k is k = 1, 2, 3, 4. As a result, the outputs of the multiplier 1118 are {Δθ (n)} / 4, {2Δθ (n)} / 4, {3Δθ (n)} / 4, and Δθ (n) (see FIG. 8).

In the present embodiment, the case where the phase rotation amount compensation is performed at the timing of the GI shown in FIG. 8 has been described. Therefore, the deterioration in the reception device (terminal) that receives the signal from the canceller device (relay device). Is preferable. However, the timing of compensation of the phase rotation amount is not limited to this. For example, the phase rotation amount of the filter coefficient may always be compensated continuously regardless of the GI.
In this case, the division unit 1117 divides the number of sample clock signals for the update interval from the phase rotation amount (for the update interval) that is the output of the addition unit 1115. Then, the multiplication unit 1118 multiplies the phase rotation amount output from the division unit 1117 by the number of times of the sample clock signal for each sample clock signal, and outputs the result to the vector signal phase rotation unit 1111. This is preferable because the canceling performance is improved by more accurately simulating the fluctuation of the transmission path.

The present invention is not limited to the first and second embodiments, and may be changed without departing from the gist of the present invention. For example, the positions of the AD converter and the DA converter may be changed.
Further, the canceller apparatus may realize the functions of the respective units by hardware such as a circuit, or may be realized by a DSP (Digital Signal Processor) or the like.

Cancellation equipment of the present invention are useful for canceling the coupling loop in the environment where the transmission path is varied.

The block diagram which shows the structural example of the wraparound cancellation apparatus which concerns on Embodiment 1 of this invention. The block diagram which shows the structural example of the coefficient update part of FIG. The block diagram which shows the structural example of the phase rotation part of FIG. The block diagram which shows the other structural example of the phase rotation part in Embodiment 1 of this invention. The block diagram which shows another structural example of the phase rotation part in Embodiment 1 of this invention. The block diagram which shows the example of a combination of the phase rotation part of FIG. 5, and a coefficient update part. The figure which shows the operation | movement timing of the wraparound cancellation in a prior art example. The figure which shows an example of the operation | movement timing of the wraparound cancellation in Embodiment 1 of this invention. The block diagram which shows the structural example of the phase rotation part of the wraparound cancellation apparatus which concerns on Embodiment 2 of this invention. The block diagram which shows the structure of the wraparound cancellation apparatus of a prior art example. Explanatory drawing which shows the change of a filter coefficient.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Rounding cancellation apparatus 101 Reception antenna 102 Reception conversion part 103 AGC
104 A / D conversion unit 105 Subtraction unit 106 FIR filter 107 Transmission path estimation unit 108 Loop estimation unit 109 Coefficient update unit 110 Delay unit 111 Phase rotation unit 112 D / A conversion unit 113 Transmission conversion unit 114 PA unit 115 Transmission antenna

Claims (5)

A sneak cancel device for removing sneak current between radio transmitting and receiving antennas of a broadcast wave relay station in digital broadcasting,
A subtracting unit that subtracts a sneak signal for canceling a sneak between transmission and reception antennas from an input signal from a reception system;
Based on the input signal after subtraction, a filter that generates a copy of the sneak signal according to a set filter coefficient and outputs the duplicate signal to the subtractor;
Based on the input signal after subtraction, a transmission path estimation unit that estimates the characteristics of the sneak path,
Based on the characteristics of the sneak path, a sneak estimation unit that estimates sneak characteristics;
A coefficient updating unit that generates and updates the filter coefficient based on the wraparound characteristic;
The phase rotation amount between the filter coefficient before update and the filter coefficient after update is calculated as a predicted value for each of the taps of the filter , the phase rotation amount is reflected in the filter coefficient after update, and the filter and the phase rotation unit to be set,
Wrapping cancel device including.   The wraparound canceling apparatus according to claim 1, wherein the phase rotation unit averages and equalizes a phase rotation amount for a predetermined number of updates. When having a plurality of symbols in the update interval of the filter coefficient,
The phase rotation unit sequentially calculates the phase rotation amount for each symbol, and calculates the phase rotation amount in the update interval;
The wraparound cancel device according to claim 1.   The wraparound cancel apparatus according to claim 3, wherein the update interval is a guide interval section of an OFDM signal.   The wraparound cancel device according to claim 3, wherein the update interval is between successive sample clock signals.

Images